DLF Spring 2009 Forum BoF Session Minutes
Notes from BoF Session, DLF Spring 2009: May 5, 2009
Note takers: Melanie Schlosser and Michelle Dalmau
In Attendance: Perry Willett (CDL), Glen Worthy (Stanford), Charles Blair (Chicago), Hugh Cayless (NYU), Kirill Fesenko (Chapel Hill), Laura Akerman (Emory), Nate Trail (LC), Cristela Garaa-Spitz (UCSD), Sally McCallum (LC), Tracy Meekleib (LC), Patrick Yott (Brown), Natasha Smith (Chapel Hill), Ceclia Tittemore (Dartsmouth), Tom Habing (UIUC), Bill Brockman (Penn State), Jim Tutlle (NCSU), Stephen Davis (Columbia), Dianne Dietrich (Cornell), Patricia Hswe (U of Illinois), Jean Godby (OCLC), David Reynolds (Johns Hopkins), Renée McBride (Chapel Hill), Aillel Arnold (NYU), Kenneth Reed (UNC Press), Chris Powell (Michigan), John Rees (Nat'l Library of Medicine), Sarah Shreeves (UIUC), Jenn Riley (IU), Kevin Hawkins (Michigan), Matthew Gibson (UVa), Melanie Schlosser (OSU) and Michelle Dalmau (IU)
- Background on TEI and the guidelines
- Introductions (~35 in attendance)
- Sign-up sheet circulating
- Clarify some things
- Reflect library realities, including different contexts, different metadata standards
- Continue to encourage standardization and machine-readability (granular metadata)
- Overview of levels, added more examples, more explanation, remove glaring errors and tag abuse, conformance
- Levels 1&2: Auto content generation, focused on fixing tag abuse, referenced an automatic workflow scenario
- Level 3: Basic structural markup, new elements (embedded text, running headers and footers), examples on notes, footnotes, etc.
- Level 4: Content analysis level, loaded with examples of different kinds of documents. New component on name tagging.
- Level 5: Projects are unique, develop specific guidelines, wanted to provide context, examples of rich semantic markup
- Were there specific use cases or tools (style sheets in mind that drove decisions to be made about the Guidelines?
- Version 2.1 reflects decision based on known software but Version 3 moves away from us. Workflows we have in mind to arrive in level 1 and level 2.
- Only design constraint is MARC record as the basis for the Header
- Specific use cases or tools trying to support? Stylesheets, etc. = Sort of. In past, decisions made based on existing DL delivery software. We strove to get away from that. We did think about workflows, but not specific stylesheets. Need consistent base of guidelines so you can build the right tools. Use cases.
- OCR workflows? = Yes. didn't flesh out all scenarios, want to flesh out workflows for future iterations. Once guidelines are stable, we will generate an ODD file for each level w/ stylesheets.
- Outsourcer scenario came up in conversation. Need to have things to point to for vendors. (TEI Tite)
- This group familiar with the most common workflows
Another customization of guidelines. Designed to be a specification a vendor could use to produce encoded text. Stripped down, removes ambiguities. Our guidelines are for things libraries do themselves. John Unsworth and Perry Trolard came up with it, then Mellon funding to survey community and determine outsourcing practices to write an informed RFP. Supposed to be a membership perk for TEI members. Cuts out startup costs in using a vendor. Enables smaller institution to aggregate content for outsourcing at low costs. Still not sure exactly how we interact with Tite, but we will be coordinating with them.
General Thoughts on the Best Practices
- Trying to go through a vendor (before Google), they didn't understand, sending them the Guidelines didn't resolve it, there were errors and issues.
- Guidelines not meant to stand alone. We link out to P5 Guidelines.
- about granularity. Compared guidelines to an existing Header. They stop at a higher level of granularity - why? Suggestion - when you don't go down to the lowest level of granularity, provide some guidance and/or say why you didn't.
- Operating constraint w/ Header - want to allow possibility of auto generation from MARC.
- Are there canonical crosswalks from TEI Header to MARC, MODS, etc.?
- No. Institutions do their own.
- Catherine Gick is working on WWP with header mapping to MARC
- Future vision: Show MARC equivalencies in element table. We need a cataloger.
- Renee McBride from Chapel Hill volunteered
- Always been troubling that TEI and EAD have elements that duplicate MARC cataloging. Need to be able to point to an external, canonical catalog record.
- We have worked hard to find a place in the Header where we can point out to other metadata. Not there yet.
- Laurent Romary pushing for linking from author element to authority file. It got approved. We will suggest something similar.
- sourceDesc: explain biblStruct instead of bibl or biblFull
- That is the intention. We should make it clearer that we are recommending this one instead of others, and give a rationale.
- Would Guidelines apply when encoding oral history transcriptions - we invented an approach of our own. Is there a place to let other people see our usage guidelines? Document use of TEI for different types of documents (i.e. non-book documents)
- Could be done on the TEI wiki (if it's not being done already). Might want to collect more use cases for future revisions of the Guidelines.
- TEI Wiki: create a space on the wiki to support use cases to develop "profiles." Useful to know which institutions are working on genre.
General SIG Issue To Take Up
- How will you communicate when not under DLF anymore? Is there a listserve?
- TEILIB-L list is main form of communication. Most of us are members of Libraries SIG under TEI. SIG meets at annual meeting in the fall. SIG deals with lots of other things, but Guidelines have been focus lately.
- Language element is missing from the Header
- We will look at it.
- Need to know use cases in the community for using METS with TEI
- Recommendation for if you are using another metadata standard to capture metadata how to recommend how much of that source metadata record to include in the TEI Header.
New Name for the Document
- use the term "profile" (as in a "METS profile")
- "Best practices for TEI in Libraries"
- "TEI Best Practices in Libraries"
- P5 has introduced 'facsimile' that replicates some of the functionality of METS. Lets you hone in on a zone in a page, coordinates - mainly for manuscript encoding. Also supports page to page linking. Weren't sure how to deal with it in a library context, where we often use METS. We introduced the @facs attribute that will let you point. We attempted to bring METS into this version of the guidelines to cover that functionality. Community wanted to know why we didn't deal with facsimile - will have to do something with it.
- Need examples of how the community supports page-level linking to facsimile
- Call for feedback, plan going forward: finalize, mappings to MARC, submit to TEI as a customization, hopefully approved and on website. Somewhere in there, we hope to get to some of the other things (working with METS, etc.). The document is ongoing. Best source of info is TEILIB email list.